Date: Mon, 28 Aug 2006 23:16:45 +0930 From: Alastair Watts <awatts@pett.com.au> To: "R. B. Riddick" <arne_woerner@yahoo.com> Cc: freebsd-geom@freebsd.org Subject: Re: gvinum behaviour on disk failure Message-ID: <44F2F3C5.7090001@pett.com.au> In-Reply-To: <20060828132846.81472.qmail@web30308.mail.mud.yahoo.com>
index | next in thread | previous in thread | raw e-mail
R. B. Riddick wrote: > Yes. Out of my theoretically modell, I have of FBSD in my brain there are at > least 2 possibilities: > > 1. use a partition/slice of the larger disk (bsdlabel... or fdisk...) Will gmirror work with slices? I thought it worked on the physical drive layer.. > and > 2. just use gmirror and hope that it build the minimum of both sizes (which it > does actually I think; so u do not have to hope very much...)... Haven't tried that, but have wondered if it will work. I guess you would have to use the smaller of the drives as the initial drive to base the mirror on. >>Yes... I noticed gvinum was very different to Greg's original vinum >>suite ;) But this seems to be a fundamental thing that's missing from >>gvinum - the ability to recover from a drive failure and to be able to >>service the request from a different plex. > > Hmm... Does gmirror overcome these drawbacks? Don't know. If it doesn't then it's still a problem in my opinion :) We have more machines using gmirror than gvinum (only 1 with gvinum), but so far only the machine with gvinum has had a drive fail since FBSD 5.3 upwards. I'd still like to hear from someone who knows the insides of gvinum as to their opinion on the earlier reported incident. Cheers, Alhome | help
Want to link to this message? Use this
URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?44F2F3C5.7090001>